• All traffic going through Ethernet interfaces (if present) on the management modules will be
interrupted during a hitless OS upgrade. This is because both management modules must be
reloaded with the new image. This applies to hitless OS upgrade only. It does not apply to hitless
switchover or failover, which does not interrupt traffic going through Ethernet interfaces on the
standby management module (the module that takes over the active role).
• Static and dynamic multi-slot trunks will flap during a hitless switchover if any of the trunk port
members reside on the management module.
• Layer 3 multicast traffic is not supported by Hitless management.
Hitless reload or switchover requirements and limitations
The section describes the design limitation on devices with the following configuration:
• 0-port management modules
• One or more third generation line cards
For hitless reload or switch-over-active-role to succeed, the following requirements and limitations
must be met:
• Hitless reload is not allowed from Switch code when the primary image is in Router code and boot
primary is a part of the configuration.
• The standby management module must be up and in an "OK {Enabled}" state.
• A configuration requiring a reload must not be pending.
• A hitless-reload must not have already been issued on the previous active management module.
• POE firmware must not be in progress.
• The SXR running configuration must not be classified as too large (greater than 512KB).
• A TFTP session must not be in progress.
• An image sync session must not be in progress.
• The current active management card cannot have a memory utilization of greater than 90% of
available memory.
• A line card hotswap must not be in progress.
If any of these conditions are not met, an appropriate error message is printed to the console and
hitless-reload or switch-over will not succeed.
With following steps, after switchover, the new standby goes into continuous reload state:
1. SXL box is running with build "x"
2. Perform
copy tftp
of build "x+1" and wait for both active and standby to sync.
3. Execute
switch-over-active-role
.
With above step, the new active comes up but the new standby tries to load the primary image "x+1"
and due to this there is image sync issue and new standby goes to continuous reload state without
recovery. Hence, it is a limitation that after
copy tftp
operation to primary,
switch-over-active-role
operation should be avoided.
What happens during a Hitless switchover or failover
This section describes the internal events that enable a controlled or forced switchover (failover) to
take place in a hitless manner, as well as the events that occur during the switchover.
Separate data and control planes
The FSX 800 and FSX 1600 management modules have separate data and control planes. The
data
plane
forwards traffic between the switch fabric modules and all of the Interface modules in the
Hitless reload or switchover requirements and limitations
126
FastIron Ethernet Switch Administration Guide
53-1003625-01